getgroupscommand.cpp

   1 /*
   2  *  getgroupscommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by westcott on 11/10/10.
   6  *  Copyright 2010 Schloss Lab. All rights reserved.
   7  *
   8  */
   9
  10 #include "getgroupscommand.h"
  11 #include "sequence.hpp"
  12 #include "listvector.hpp"
  13 #include "sharedutilities.h"
  14 #include "inputdata.h"
  15
  16 //**********************************************************************************************************************
  17 vector<string> GetGroupsCommand::setParameters(){
  18         try {
  19                 CommandParameter pfasta("fasta", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pfasta);
  20                 CommandParameter pshared("shared", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pshared);
  21                 CommandParameter pname("name", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(pname);
  22                 CommandParameter pgroup("group", "InputTypes", "", "", "none", "FNGLT-sharedGroup", "none",false,false); parameters.push_back(pgroup);
  23                 CommandParameter plist("list", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(plist);
  24                 CommandParameter ptaxonomy("taxonomy", "InputTypes", "", "", "none", "FNGLT", "none",false,false); parameters.push_back(ptaxonomy);
  25                 CommandParameter paccnos("accnos", "InputTypes", "", "", "none", "none", "none",false,false); parameters.push_back(paccnos);
  26                 CommandParameter pgroups("groups", "String", "", "", "", "", "",false,false); parameters.push_back(pgroups);
  27                 CommandParameter pinputdir("inputdir", "String", "", "", "", "", "",false,false); parameters.push_back(pinputdir);
  28                 CommandParameter poutputdir("outputdir", "String", "", "", "", "", "",false,false); parameters.push_back(poutputdir);
  29
  30                 vector<string> myArray;
  31                 for (int i = 0; i < parameters.size(); i++) {   myArray.push_back(parameters[i].name);          }
  32                 return myArray;
  33         }
  34         catch(exception& e) {
  35                 m->errorOut(e, "GetGroupsCommand", "setParameters");
  36                 exit(1);
  37         }
  38 }
  39 //**********************************************************************************************************************
  40 string GetGroupsCommand::getHelpString(){
  41         try {
  42                 string helpString = "";
  43                 helpString += "The get.groups command selects sequences from a specfic group or set of groups from the following file types: fasta, name, group, list, taxonomy or shared file.\n";
  44                 helpString += "It outputs a file containing the sequences in the those specified groups, or a sharedfile containing only those groups.\n";
  45                 helpString += "The get.groups command parameters are accnos, fasta, name, group, list, taxonomy, shared and groups. The group parameter is required, unless you have a current group file, or are using a shared file.\n";
  46                 helpString += "You must also provide an accnos containing the list of groups to get or set the groups parameter to the groups you wish to select.\n";
  47                 helpString += "The groups parameter allows you to specify which of the groups in your groupfile you would like.  You can separate group names with dashes.\n";
  48                 helpString += "The get.groups command should be in the following format: get.groups(accnos=yourAccnos, fasta=yourFasta, group=yourGroupFile).\n";
  49                 helpString += "Example get.groups(accnos=amazon.accnos, fasta=amazon.fasta, group=amazon.groups).\n";
  50                 helpString += "or get.groups(groups=pasture, fasta=amazon.fasta, group=amazon.groups).\n";
  51                 helpString += "Note: No spaces between parameter labels (i.e. fasta), '=' and parameters (i.e.yourFasta).\n";
  52                 return helpString;
  53         }
  54         catch(exception& e) {
  55                 m->errorOut(e, "GetGroupsCommand", "getHelpString");
  56                 exit(1);
  57         }
  58 }
  59 //**********************************************************************************************************************
  60 GetGroupsCommand::GetGroupsCommand(){
  61         try {
  62                 abort = true; calledHelp = true;
  63                 setParameters();
  64                 vector<string> tempOutNames;
  65                 outputTypes["fasta"] = tempOutNames;
  66                 outputTypes["taxonomy"] = tempOutNames;
  67                 outputTypes["name"] = tempOutNames;
  68                 outputTypes["group"] = tempOutNames;
  69                 outputTypes["list"] = tempOutNames;
  70                 outputTypes["shared"] = tempOutNames;
  71         }
  72         catch(exception& e) {
  73                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
  74                 exit(1);
  75         }
  76 }
  77 //**********************************************************************************************************************
  78 GetGroupsCommand::GetGroupsCommand(string option)  {
  79         try {
  80                 abort = false; calledHelp = false;
  81
  82                 //allow user to run help
  83                 if(option == "help") { help(); abort = true; calledHelp = true; }
  84                 else if(option == "citation") { citation(); abort = true; calledHelp = true;}
  85
  86                 else {
  87                         vector<string> myArray = setParameters();
  88
  89                         OptionParser parser(option);
  90                         map<string,string> parameters = parser.getParameters();
  91
  92                         ValidParameters validParameter;
  93                         map<string,string>::iterator it;
  94
  95                         //check to make sure all parameters are valid for command
  96                         for (it = parameters.begin(); it != parameters.end(); it++) {
  97                                 if (validParameter.isValidParameter(it->first, myArray, it->second) != true) {  abort = true;  }
  98                         }
  99
 100                         //initialize outputTypes
 101                         vector<string> tempOutNames;
 102                         outputTypes["fasta"] = tempOutNames;
 103                         outputTypes["taxonomy"] = tempOutNames;
 104                         outputTypes["name"] = tempOutNames;
 105                         outputTypes["group"] = tempOutNames;
 106                         outputTypes["list"] = tempOutNames;
 107                         outputTypes["shared"] = tempOutNames;
 108
 109
 110                         //if the user changes the output directory command factory will send this info to us in the output parameter
 111                         outputDir = validParameter.validFile(parameters, "outputdir", false);           if (outputDir == "not found"){  outputDir = "";         }
 112
 113                         //if the user changes the input directory command factory will send this info to us in the output parameter
 114                         string inputDir = validParameter.validFile(parameters, "inputdir", false);
 115                         if (inputDir == "not found"){   inputDir = "";          }
 116                         else {
 117                                 string path;
 118                                 it = parameters.find("fasta");
 119                                 //user has given a template file
 120                                 if(it != parameters.end()){
 121                                         path = m->hasPath(it->second);
 122                                         //if the user has not given a path then, add inputdir. else leave path alone.
 123                                         if (path == "") {       parameters["fasta"] = inputDir + it->second;            }
 124                                 }
 125
 126                                 it = parameters.find("accnos");
 127                                 //user has given a template file
 128                                 if(it != parameters.end()){
 129                                         path = m->hasPath(it->second);
 130                                         //if the user has not given a path then, add inputdir. else leave path alone.
 131                                         if (path == "") {       parameters["accnos"] = inputDir + it->second;           }
 132                                 }
 133
 134                                 it = parameters.find("list");
 135                                 //user has given a template file
 136                                 if(it != parameters.end()){
 137                                         path = m->hasPath(it->second);
 138                                         //if the user has not given a path then, add inputdir. else leave path alone.
 139                                         if (path == "") {       parameters["list"] = inputDir + it->second;             }
 140                                 }
 141
 142                                 it = parameters.find("name");
 143                                 //user has given a template file
 144                                 if(it != parameters.end()){
 145                                         path = m->hasPath(it->second);
 146                                         //if the user has not given a path then, add inputdir. else leave path alone.
 147                                         if (path == "") {       parameters["name"] = inputDir + it->second;             }
 148                                 }
 149
 150                                 it = parameters.find("group");
 151                                 //user has given a template file
 152                                 if(it != parameters.end()){
 153                                         path = m->hasPath(it->second);
 154                                         //if the user has not given a path then, add inputdir. else leave path alone.
 155                                         if (path == "") {       parameters["group"] = inputDir + it->second;            }
 156                                 }
 157
 158                                 it = parameters.find("taxonomy");
 159                                 //user has given a template file
 160                                 if(it != parameters.end()){
 161                                         path = m->hasPath(it->second);
 162                                         //if the user has not given a path then, add inputdir. else leave path alone.
 163                                         if (path == "") {       parameters["taxonomy"] = inputDir + it->second;         }
 164                                 }
 165
 166                                 it = parameters.find("shared");
 167                                 //user has given a template file
 168                                 if(it != parameters.end()){
 169                                         path = m->hasPath(it->second);
 170                                         //if the user has not given a path then, add inputdir. else leave path alone.
 171                                         if (path == "") {       parameters["shared"] = inputDir + it->second;           }
 172                                 }
 173                         }
 174
 175
 176                         //check for required parameters
 177                         accnosfile = validParameter.validFile(parameters, "accnos", true);
 178                         if (accnosfile == "not open") { abort = true; }
 179                         else if (accnosfile == "not found") {  accnosfile = ""; }
 180                         else { m->setAccnosFile(accnosfile); }
 181
 182                         fastafile = validParameter.validFile(parameters, "fasta", true);
 183                         if (fastafile == "not open") { abort = true; }
 184                         else if (fastafile == "not found") {  fastafile = "";  }
 185                         else { m->setFastaFile(fastafile); }
 186
 187                         namefile = validParameter.validFile(parameters, "name", true);
 188                         if (namefile == "not open") { abort = true; }
 189                         else if (namefile == "not found") {  namefile = "";  }
 190                         else { m->setNameFile(namefile); }
 191
 192                         groupfile = validParameter.validFile(parameters, "group", true);
 193                         if (groupfile == "not open") { abort = true; }
 194                         else if (groupfile == "not found") {
 195                                 //if there is a current group file, use it
 196                                 groupfile = m->getGroupFile();
 197                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 198                                 else {  m->mothurOut("You have no current groupfile and the group parameter is required."); m->mothurOutEndLine(); abort = true; }
 199                         }else { m->setGroupFile(groupfile); }
 200
 201                         listfile = validParameter.validFile(parameters, "list", true);
 202                         if (listfile == "not open") { abort = true; }
 203                         else if (listfile == "not found") {  listfile = "";  }
 204                         else { m->setListFile(listfile); }
 205
 206                         taxfile = validParameter.validFile(parameters, "taxonomy", true);
 207                         if (taxfile == "not open") { abort = true; }
 208                         else if (taxfile == "not found") {  taxfile = "";  }
 209                         else { m->setTaxonomyFile(taxfile); }
 210
 211                         groups = validParameter.validFile(parameters, "groups", false);
 212                         if (groups == "not found") { groups = ""; }
 213                         else {
 214                                 m->splitAtDash(groups, Groups);
 215                                 m->setGroups(Groups);
 216                         }
 217
 218                         sharedfile = validParameter.validFile(parameters, "shared", true);
 219                         if (sharedfile == "not open") { sharedfile = ""; abort = true; }
 220                         else if (sharedfile == "not found") {  sharedfile = "";  }
 221                         else { m->setSharedFile(sharedfile); }
 222
 223                         groupfile = validParameter.validFile(parameters, "group", true);
 224                         if (groupfile == "not open") { groupfile = ""; abort = true; }
 225                         else if (groupfile == "not found") {    groupfile = ""; }
 226                         else { m->setGroupFile(groupfile); }
 227
 228                         if ((sharedfile == "") && (groupfile == "")) {
 229                                 //is there are current file available for any of these?
 230                                 if ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")) {
 231                                         //give priority to group, then shared
 232                                         groupfile = m->getGroupFile();
 233                                         if (groupfile != "") {  m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 234                                         else {
 235                                                 sharedfile = m->getSharedFile();
 236                                                 if (sharedfile != "") { m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 237                                                 else {
 238                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 239                                                 }
 240                                         }
 241                                 }else {
 242                                         //give priority to shared, then group
 243                                         sharedfile = m->getSharedFile();
 244                                         if (sharedfile != "") {  m->mothurOut("Using " + sharedfile + " as input file for the shared parameter."); m->mothurOutEndLine(); }
 245                                         else {
 246                                                 groupfile = m->getGroupFile();
 247                                                 if (groupfile != "") { m->mothurOut("Using " + groupfile + " as input file for the group parameter."); m->mothurOutEndLine(); }
 248                                                 else {
 249                                                         m->mothurOut("You have no current groupfile or sharedfile and one is required."); m->mothurOutEndLine(); abort = true;
 250                                                 }
 251                                         }
 252                                 }
 253                         }
 254
 255                         if ((accnosfile == "") && (Groups.size() == 0)) { m->mothurOut("You must provide an accnos file or specify groups using the groups parameter."); m->mothurOutEndLine(); abort = true; }
 256
 257                         if ((fastafile == "") && (namefile == "") && (groupfile == "")  && (sharedfile == "") && (listfile == "") && (taxfile == ""))  { m->mothurOut("You must provide at least one of the following: fasta, name, taxonomy, group, shared or list."); m->mothurOutEndLine(); abort = true; }
 258                         if ((groupfile == "") && ((namefile != "") || (fastafile != "") || (listfile != "") || (taxfile != "")))  { m->mothurOut("If using a fasta, name, taxonomy, group or list, then you must provide a group file."); m->mothurOutEndLine(); abort = true; }
 259
 260                 }
 261
 262         }
 263         catch(exception& e) {
 264                 m->errorOut(e, "GetGroupsCommand", "GetGroupsCommand");
 265                 exit(1);
 266         }
 267 }
 268 //**********************************************************************************************************************
 269
 270 int GetGroupsCommand::execute(){
 271         try {
 272
 273                 if (abort == true) { if (calledHelp) { return 0; }  return 2;   }
 274
 275                 //get groups you want to remove
 276                 if (accnosfile != "") { readAccnos(); }
 277
 278                 if (groupfile != "") {
 279                         groupMap = new GroupMap(groupfile);
 280                         groupMap->readMap();
 281
 282                         //make sure groups are valid
 283                         //takes care of user setting groupNames that are invalid or setting groups=all
 284                         SharedUtil* util = new SharedUtil();
 285                         vector<string> gNamesOfGroups = groupMap->getNamesOfGroups();
 286                         util->setGroups(Groups, gNamesOfGroups);
 287                         groupMap->setNamesOfGroups(gNamesOfGroups);
 288                         delete util;
 289
 290                         //fill names with names of sequences that are from the groups we want to remove
 291                         fillNames();
 292
 293                         delete groupMap;
 294                 }
 295
 296                 if (m->control_pressed) { return 0; }
 297
 298                 //read through the correct file and output lines you want to keep
 299                 if (namefile != "")                     {               readName();             }
 300                 if (fastafile != "")            {               readFasta();    }
 301                 if (groupfile != "")            {               readGroup();    }
 302                 if (listfile != "")                     {               readList();             }
 303                 if (taxfile != "")                      {               readTax();              }
 304                 if (sharedfile != "")           {               readShared();   }
 305
 306                 if (m->control_pressed) { for (int i = 0; i < outputNames.size(); i++) {        m->mothurRemove(outputNames[i]); } return 0; }
 307
 308
 309                 if (outputNames.size() != 0) {
 310                         m->mothurOutEndLine();
 311                         m->mothurOut("Output File names: "); m->mothurOutEndLine();
 312                         for (int i = 0; i < outputNames.size(); i++) {  m->mothurOut(outputNames[i]); m->mothurOutEndLine();    }
 313                         m->mothurOutEndLine();
 314
 315                         //set fasta file as new current fastafile
 316                         string current = "";
 317                         itTypes = outputTypes.find("fasta");
 318                         if (itTypes != outputTypes.end()) {
 319                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setFastaFile(current); }
 320                         }
 321
 322                         itTypes = outputTypes.find("name");
 323                         if (itTypes != outputTypes.end()) {
 324                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setNameFile(current); }
 325                         }
 326
 327                         itTypes = outputTypes.find("group");
 328                         if (itTypes != outputTypes.end()) {
 329                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setGroupFile(current); }
 330                         }
 331
 332                         itTypes = outputTypes.find("list");
 333                         if (itTypes != outputTypes.end()) {
 334                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setListFile(current); }
 335                         }
 336
 337                         itTypes = outputTypes.find("taxonomy");
 338                         if (itTypes != outputTypes.end()) {
 339                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setTaxonomyFile(current); }
 340                         }
 341
 342                         itTypes = outputTypes.find("shared");
 343                         if (itTypes != outputTypes.end()) {
 344                                 if ((itTypes->second).size() != 0) { current = (itTypes->second)[0]; m->setSharedFile(current); }
 345                         }
 346                 }
 347
 348                 return 0;
 349         }
 350
 351         catch(exception& e) {
 352                 m->errorOut(e, "GetGroupsCommand", "execute");
 353                 exit(1);
 354         }
 355 }
 356
 357 //**********************************************************************************************************************
 358 int GetGroupsCommand::readFasta(){
 359         try {
 360                 string thisOutputDir = outputDir;
 361                 if (outputDir == "") {  thisOutputDir += m->hasPath(fastafile);  }
 362                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(fastafile)) + "pick" + m->getExtension(fastafile);
 363
 364                 ofstream out;
 365                 m->openOutputFile(outputFileName, out);
 366
 367                 ifstream in;
 368                 m->openInputFile(fastafile, in);
 369                 string name;
 370
 371                 bool wroteSomething = false;
 372                 int selectedCount = 0;
 373
 374                 while(!in.eof()){
 375                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 376
 377                         Sequence currSeq(in);
 378                         name = currSeq.getName();
 379
 380                         if (name != "") {
 381                                 //if this name is in the accnos file
 382                                 if (names.count(name) != 0) {
 383                                         wroteSomething = true;
 384
 385                                         currSeq.printSequence(out);
 386                                         selectedCount++;
 387                                 }else{
 388                                         //if you are not in the accnos file check if you are a name that needs to be changed
 389                                         map<string, string>::iterator it = uniqueToRedundant.find(name);
 390                                         if (it != uniqueToRedundant.end()) {
 391                                                 wroteSomething = true;
 392                                                 currSeq.setName(it->second);
 393                                                 currSeq.printSequence(out);
 394                                                 selectedCount++;
 395                                         }
 396                                 }
 397                         }
 398                         m->gobble(in);
 399                 }
 400                 in.close();
 401                 out.close();
 402
 403                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 404                 outputTypes["fasta"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 405
 406                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your fasta file."); m->mothurOutEndLine();
 407
 408
 409                 return 0;
 410
 411         }
 412         catch(exception& e) {
 413                 m->errorOut(e, "GetGroupsCommand", "readFasta");
 414                 exit(1);
 415         }
 416 }
 417 //**********************************************************************************************************************
 418 int GetGroupsCommand::readShared(){
 419         try {
 420                 string thisOutputDir = outputDir;
 421                 if (outputDir == "") {  thisOutputDir += m->hasPath(sharedfile);  }
 422
 423                 InputData input(sharedfile, "sharedfile");
 424                 vector<SharedRAbundVector*> lookup = input.getSharedRAbundVectors();
 425
 426                 bool wroteSomething = false;
 427
 428                 while(lookup[0] != NULL) {
 429
 430                         string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(sharedfile)) + lookup[0]->getLabel() + ".pick" + m->getExtension(sharedfile);
 431                         ofstream out;
 432                         m->openOutputFile(outputFileName, out);
 433                         outputTypes["shared"].push_back(outputFileName);  outputNames.push_back(outputFileName);
 434
 435                         if (m->control_pressed) { out.close();  m->mothurRemove(outputFileName);  for (int i = 0; i < lookup.size(); i++) { delete lookup[i]; } return 0; }
 436
 437                         lookup[0]->printHeaders(out);
 438
 439                         for (int i = 0; i < lookup.size(); i++) {
 440                                 out << lookup[i]->getLabel() << '\t' << lookup[i]->getGroup() << '\t';
 441                                 lookup[i]->print(out);
 442                                 wroteSomething = true;
 443
 444                         }
 445
 446                         //get next line to process
 447                         //prevent memory leak
 448                         for (int i = 0; i < lookup.size(); i++) {  delete lookup[i];  }
 449                         lookup = input.getSharedRAbundVectors();
 450
 451                         out.close();
 452                 }
 453
 454                 if (wroteSomething == false) {  m->mothurOut("Your file contains only the groups you wish to remove."); m->mothurOutEndLine();  }
 455
 456                 string groupsString = "";
 457                 for (int i = 0; i < Groups.size()-1; i++) {     groupsString += Groups[i] + ", "; }
 458                 groupsString += Groups[Groups.size()-1];
 459
 460                 m->mothurOut("Selected groups: " + groupsString + " from your shared file."); m->mothurOutEndLine();
 461
 462                 return 0;
 463
 464         }
 465         catch(exception& e) {
 466                 m->errorOut(e, "GetGroupsCommand", "readShared");
 467                 exit(1);
 468         }
 469 }
 470 //**********************************************************************************************************************
 471 int GetGroupsCommand::readList(){
 472         try {
 473                 string thisOutputDir = outputDir;
 474                 if (outputDir == "") {  thisOutputDir += m->hasPath(listfile);  }
 475                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(listfile)) + "pick" +  m->getExtension(listfile);
 476
 477                 ofstream out;
 478                 m->openOutputFile(outputFileName, out);
 479
 480                 ifstream in;
 481                 m->openInputFile(listfile, in);
 482
 483                 bool wroteSomething = false;
 484                 int selectedCount = 0;
 485
 486                 while(!in.eof()){
 487
 488                         selectedCount = 0;
 489
 490                         //read in list vector
 491                         ListVector list(in);
 492
 493                         //make a new list vector
 494                         ListVector newList;
 495                         newList.setLabel(list.getLabel());
 496
 497                         //for each bin
 498                         for (int i = 0; i < list.getNumBins(); i++) {
 499                                 if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 500
 501                                 //parse out names that are in accnos file
 502                                 string binnames = list.get(i);
 503
 504                                 string newNames = "";
 505                                 while (binnames.find_first_of(',') != -1) {
 506                                         string name = binnames.substr(0,binnames.find_first_of(','));
 507                                         binnames = binnames.substr(binnames.find_first_of(',')+1, binnames.length());
 508
 509                                         //if that name is in the .accnos file, add it
 510                                         if (names.count(name) != 0) {  newNames += name + ",";  selectedCount++;  }
 511                                         else{
 512                                                 //if you are not in the accnos file check if you are a name that needs to be changed
 513                                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 514                                                 if (it != uniqueToRedundant.end()) {
 515                                                         newNames += it->second + ",";
 516                                                         selectedCount++;
 517                                                 }
 518                                         }
 519                                 }
 520
 521                                 //get last name
 522                                 if (names.count(binnames) != 0) {  newNames += binnames + ",";  selectedCount++;  }
 523                                 else{
 524                                         //if you are not in the accnos file check if you are a name that needs to be changed
 525                                         map<string, string>::iterator it = uniqueToRedundant.find(binnames);
 526                                         if (it != uniqueToRedundant.end()) {
 527                                                 newNames += it->second + ",";
 528                                                 selectedCount++;
 529                                         }
 530                                 }
 531
 532                                 //if there are names in this bin add to new list
 533                                 if (newNames != "") {
 534                                         newNames = newNames.substr(0, newNames.length()-1); //rip off extra comma
 535                                         newList.push_back(newNames);
 536                                 }
 537                         }
 538
 539                         //print new listvector
 540                         if (newList.getNumBins() != 0) {
 541                                 wroteSomething = true;
 542                                 newList.print(out);
 543                         }
 544
 545                         m->gobble(in);
 546                 }
 547                 in.close();
 548                 out.close();
 549
 550                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 551                 outputTypes["list"].push_back(outputFileName); outputNames.push_back(outputFileName);
 552
 553                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your list file."); m->mothurOutEndLine();
 554
 555                 return 0;
 556
 557         }
 558         catch(exception& e) {
 559                 m->errorOut(e, "GetGroupsCommand", "readList");
 560                 exit(1);
 561         }
 562 }
 563 //**********************************************************************************************************************
 564 int GetGroupsCommand::readName(){
 565         try {
 566                 string thisOutputDir = outputDir;
 567                 if (outputDir == "") {  thisOutputDir += m->hasPath(namefile);  }
 568                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(namefile)) + "pick" + m->getExtension(namefile);
 569
 570                 ofstream out;
 571                 m->openOutputFile(outputFileName, out);
 572
 573                 ifstream in;
 574                 m->openInputFile(namefile, in);
 575                 string name, firstCol, secondCol;
 576
 577                 bool wroteSomething = false;
 578                 int selectedCount = 0;
 579
 580                 while(!in.eof()){
 581                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 582
 583                         in >> firstCol;         m->gobble(in);
 584                         in >> secondCol;
 585
 586                         vector<string> parsedNames;
 587                         m->splitAtComma(secondCol, parsedNames);
 588
 589                         vector<string> validSecond;  validSecond.clear();
 590                         for (int i = 0; i < parsedNames.size(); i++) {
 591                                 if (names.count(parsedNames[i]) != 0) {
 592                                         validSecond.push_back(parsedNames[i]);
 593                                 }
 594                         }
 595
 596                         selectedCount += validSecond.size();
 597
 598                         //if the name in the first column is in the set then print it and any other names in second column also in set
 599                         if (names.count(firstCol) != 0) {
 600
 601                                 wroteSomething = true;
 602
 603                                 out << firstCol << '\t';
 604
 605                                 //you know you have at least one valid second since first column is valid
 606                                 for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 607                                 out << validSecond[validSecond.size()-1] << endl;
 608
 609                                 //make first name in set you come to first column and then add the remaining names to second column
 610                         }else {
 611
 612                                 //you want part of this row
 613                                 if (validSecond.size() != 0) {
 614
 615                                         wroteSomething = true;
 616
 617                                         out << validSecond[0] << '\t';
 618
 619                                         //you know you have at least one valid second since first column is valid
 620                                         for (int i = 0; i < validSecond.size()-1; i++) {  out << validSecond[i] << ',';  }
 621                                         out << validSecond[validSecond.size()-1] << endl;
 622                                         uniqueToRedundant[firstCol] = validSecond[0];
 623                                 }
 624                         }
 625
 626                         m->gobble(in);
 627                 }
 628                 in.close();
 629                 out.close();
 630
 631                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 632                 outputTypes["name"].push_back(outputFileName); outputNames.push_back(outputFileName);
 633
 634                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your name file."); m->mothurOutEndLine();
 635
 636                 return 0;
 637         }
 638         catch(exception& e) {
 639                 m->errorOut(e, "GetGroupsCommand", "readName");
 640                 exit(1);
 641         }
 642 }
 643
 644 //**********************************************************************************************************************
 645 int GetGroupsCommand::readGroup(){
 646         try {
 647                 string thisOutputDir = outputDir;
 648                 if (outputDir == "") {  thisOutputDir += m->hasPath(groupfile);  }
 649                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(groupfile)) + "pick" + m->getExtension(groupfile);
 650
 651                 ofstream out;
 652                 m->openOutputFile(outputFileName, out);
 653
 654                 ifstream in;
 655                 m->openInputFile(groupfile, in);
 656                 string name, group;
 657
 658                 bool wroteSomething = false;
 659                 int selectedCount = 0;
 660
 661                 while(!in.eof()){
 662                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 663
 664                         in >> name;                             //read from first column
 665                         in >> group;                    //read from second column
 666
 667                         //if this name is in the accnos file
 668                         if (names.count(name) != 0) {
 669                                 wroteSomething = true;
 670                                 out << name << '\t' << group << endl;
 671                                 selectedCount++;
 672                         }
 673
 674                         m->gobble(in);
 675                 }
 676                 in.close();
 677                 out.close();
 678
 679                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 680                 outputTypes["group"].push_back(outputFileName); outputNames.push_back(outputFileName);
 681
 682                 m->mothurOut("Selected " + toString(selectedCount) + " sequences from your group file."); m->mothurOutEndLine();
 683
 684                 return 0;
 685         }
 686         catch(exception& e) {
 687                 m->errorOut(e, "GetGroupsCommand", "readGroup");
 688                 exit(1);
 689         }
 690 }
 691 //**********************************************************************************************************************
 692 int GetGroupsCommand::readTax(){
 693         try {
 694                 string thisOutputDir = outputDir;
 695                 if (outputDir == "") {  thisOutputDir += m->hasPath(taxfile);  }
 696                 string outputFileName = thisOutputDir + m->getRootName(m->getSimpleName(taxfile)) + "pick" + m->getExtension(taxfile);
 697                 ofstream out;
 698                 m->openOutputFile(outputFileName, out);
 699
 700                 ifstream in;
 701                 m->openInputFile(taxfile, in);
 702                 string name, tax;
 703
 704                 bool wroteSomething = false;
 705
 706                 while(!in.eof()){
 707                         if (m->control_pressed) { in.close();  out.close();  m->mothurRemove(outputFileName);  return 0; }
 708
 709                         in >> name;                             //read from first column
 710                         in >> tax;                      //read from second column
 711
 712                         //if this name is in the accnos file
 713                         if (names.count(name) != 0) {
 714                                 wroteSomething = true;
 715                                 out << name << '\t' << tax << endl;
 716                         }else{
 717                                 //if you are not in the accnos file check if you are a name that needs to be changed
 718                                 map<string, string>::iterator it = uniqueToRedundant.find(name);
 719                                 if (it != uniqueToRedundant.end()) {
 720                                         wroteSomething = true;
 721                                         out << it->second << '\t' << tax << endl;
 722                                 }
 723                         }
 724
 725                         m->gobble(in);
 726                 }
 727                 in.close();
 728                 out.close();
 729
 730                 if (wroteSomething == false) {  m->mothurOut("Your file does NOT contain sequences from the groups you wish to get."); m->mothurOutEndLine();  }
 731                 outputTypes["taxonomy"].push_back(outputFileName); outputNames.push_back(outputFileName);
 732
 733                 return 0;
 734         }
 735         catch(exception& e) {
 736                 m->errorOut(e, "GetGroupsCommand", "readTax");
 737                 exit(1);
 738         }
 739 }
 740 //**********************************************************************************************************************
 741 void GetGroupsCommand::readAccnos(){
 742         try {
 743                 Groups.clear();
 744
 745                 ifstream in;
 746                 m->openInputFile(accnosfile, in);
 747                 string name;
 748
 749                 while(!in.eof()){
 750                         in >> name;
 751
 752                         Groups.push_back(name);
 753
 754                         m->gobble(in);
 755                 }
 756                 in.close();
 757
 758                 m->setGroups(Groups);
 759
 760         }
 761         catch(exception& e) {
 762                 m->errorOut(e, "GetGroupsCommand", "readAccnos");
 763                 exit(1);
 764         }
 765 }
 766 //**********************************************************************************************************************
 767 int GetGroupsCommand::fillNames(){
 768         try {
 769                 vector<string> seqs = groupMap->getNamesSeqs();
 770
 771                 for (int i = 0; i < seqs.size(); i++) {
 772
 773                         if (m->control_pressed) { return 0; }
 774
 775                         string group = groupMap->getGroup(seqs[i]);
 776
 777                         if (m->inUsersGroups(group, Groups)) {
 778                                 names.insert(seqs[i]);
 779                         }
 780                 }
 781
 782                 return 0;
 783         }
 784         catch(exception& e) {
 785                 m->errorOut(e, "GetGroupsCommand", "fillNames");
 786                 exit(1);
 787         }
 788 }
 789
 790 //**********************************************************************************************************************
 791
 792